Decentralized Safe Control for Multi-Robot Navigation in Dynamic Environments with Limited Sensing

Our research addresses the challenging multi-agent safe control problem where agents must reach their goals while avoiding collisions. Avoidance constraints are enforced within a limited sensing field, adding practical relevance to the problem. We propose a novel approach based on tractable Control Lyapunov Function (CLF)based Quadratic Programs (QPs) for individual agents, enabling goal tracking while considering the dynamics of the obstacles in their limited sensing range. Our framework is highly adaptable, accommodating a large number of agents and ensuring scalability. Extensive experiments with differential drive robots illustrate the computational efficiency and scalability of our approach, even in highly occluded environments with large number of robots.